The Haiku: why I personally feel the 3S lights have surpassed the PD lights

[nbp]

We could always do a swap for a week. Maybe. Just a thought.

Right, but then I have to give it back. :devil:

I have experience with the 119 as I had one briefly in a Ti Mule but then I traded that LE to someone who wanted one and got a neutral XP-G LE out of it which is now in an EN Mule. Now I have a hankering for a hi cri Haiku. :naughty:

In regards to this blinky business, I still have not seen the PWM people complained about in the old LEs, so as long as most customers are like me, you'll probably have no problems Don with the new ones as we likely won't notice.

 

[McGizmo]

Re: The Haiku: why the 3S McClickie lights have surpassed the PD lights!

Don, based on your description of the intermittent sinewave pattern, it does sound like the circuit is periodically re-testing a range of voltages to determine the ideal one for the current conditions. There is a similar behavior in the GDuP drivers I installed in a few Lunasol lights, where significant changes in ambient temperature will cause the light to blink, adjust its brightness, and blink again the next time it's turned on. It doesn't re-test itself repeatedly while on, though, only once at the beginning. It's actually an incredibly useful feature and it allows me to install the drivers into a light regardless of which kind of emitter it has; if I could reprogram the current settings on my own, I'd just buy a batch of the drivers and never worry about finding a good driver again, just keep re-using the ones I have.

Most likely the drivers you already have could be flashed with new software to remove the repeated re-testing if desired, or at least reduce the frequency to once a minute or something else more reasonable than once every few seconds, but only your vendor can say for sure.

fyrstormer,

I believe you are on the right track here and my understanding is that these converters are designed to allow for a variation in output current up to 10 mA. When the current exceeds 10 mA of current beyond the target level it steps in and corrects the condition. What can be seen with the XM-L is this correction process. To use probably a poor analogy, it's akin to a speed governor kicking it. I personally consider this quirk as an artifact of no significant consequence but am compelled to let it be known that it is there now that I am aware of its presence. Initially these converters were not programmed to evaluate a change in current to the LED as the Vf dropped by virtue of heating up and I saw the results of this on my bench and was able to measure the current to the LED climb significantly as it heated up. Now this was not apparent to the naked eye but something potentially detrimental to the LED and required addressing. If anything seeing the light level pulse is an indicator that the converter is monitoring the current to the LED and keeping it within design limits. Seeing this pulse in intensity is an indication that the converter is alive and well, as it were. I'm going to step away for a quick bench test...

OK, I'm back. What I did was set my bench supply with a current limit at 600 mA and the supply voltage at 4VCD. I hooked up the bench supply to a XM-L mounted on a MCPCB. On initial fire up the bench supply was driving the LED at 600 mA with a Vin of 2.9 V. That is to say that this particular LED at room temperature has a Vf of 2.9 at 600 mA. So then I set the bench supply at 2.9 volts and went ahead and set the current limit at 1.5 amps. I attached the LED to the bench supply and watched the current to the LED grow from 600 mA to 800 mA, 37 seconds later. I shut the power off as I had seen enough.

It is important to consider the fact that this test LED had only the MCPCB as a heat sink so it heated up quicker than it would with the MCPCB integrated into a Light engine and housed in a flashlight but the fact remains that the XM-L, Vf is significantly dependent on temperature as well as current. In a flashlight it might take minutes and not 37 seconds for this drift in Vf but the drift is still there and the greater you allow the current to increase the greater the difference in apparent output would be when it is brought back in line. It is my suspicion that other LED's may not have Vf drift as a function of temperature like the XM-L. Since the Vf drops as the XM-L heats up having this monitored and corrected is a good idea, IMHO and it's also cool that as the LED heats up, its resistance drops making it more efficient at generating light as it requires less voltage to maintain the same current level.

If the designer of the 3S converter can come up with a program that will monitor the current to the LED in a less obvious manner than I will be receptive to a program change and implement such a change in a reasonable manner. I will also let you guys know if any change or revision comes about. But in the mean time I plan to go forward with what I have on hand and with the disclosure that there is now a known issue (artifact) in the sense of an apparent pulse type change in flux when the converter steps in and corrects for a temperature induced drift in LED Vf. To be clear, I have no plans on dismantling any existing light engines should a new program be developed because I don't think the fix warrants the surgery or shipping to and back from Japan.

 

[fyrstormer]

Re: The Haiku: why the 3S McClickie lights have surpassed the PD lights!

If the converter is in fact periodically recalibrating itself, perhaps the programming can be adjusted to perform the recalibration more slowly, and to start with an initial reading and then ramp the voltage up or down only in the direction that brings the current flow closer to the correct amount, instead of swinging above and below as it seems to do at present. That's how the GDuP behaves, so I have to imagine a similarly-complex circuit could be made to behave similarly as well -- minus the "I'm re-calibrating myself now" warning flashes at the beginning and end of the process.

 

[js]

Re: The Haiku: why the 3S McClickie lights have surpassed the PD lights!

. . .
Most likely the drivers you already have could be flashed with new software to remove the repeated re-testing if desired, or at least reduce the frequency to once a minute or something else more reasonable than once every few seconds, but only your vendor can say for sure.

Don, I was under the impression, from reading your post, that this correction was not happening as often as "once every few seconds." This is the relevant bit:

In graphing the flux on the y axis you can see basically a straight line and then a time segment where there is a flat sine wave repeating a number of times and then back to straight.

i.e. when it is correcting, it adjusts a few times, with a heart-beat type pulse, each "beat" being a 1/3 of a second, but then it goes back to a straight line for however long. It's not like it's beating/pulsing every few seconds indefinitely, right?

Please clarify, if you would.

 

[wquiles]

Re: The Haiku: why the 3S McClickie lights have surpassed the PD lights!

It would not be recalibration - that does not make sense to me from the strict meaning of what calibration means to me: to adjust the output and/or multipliers to match a specific/known voltage/current point, in order to minimize error/difference from part to part/circuit to circuit against the "golden standard" or theoretical value.

However, the behavior you guys describe here does very likely sound like normal behavior (in general terms) to what any/all regulating circuits do. In this case (I am assuming it is a current regulated driver) the circuit is attempting to maintain a set current level, as the battery slowly drains, and as the vf of the LED changes (as Don stated above). But the regulator has some intrinsic step-size used to adjust the output voltage up or down to maintain the constant output current.

My best educated guess based on my own experience with regulated drivers is that what you guys are seeing is the normal oscillation of the algorithm around a target output. The algorithm is going one step up and one step down, as it continuously adjusts the output voltage. Of course not knowing more about the driver makes it harder to guess if the step is hardware controlled or software controlled (as with my own regulator, but even then I am limited to the micro-controller that I am using), so it is hard to say if the step size can be adjusted, and/or how frequently/slowly this can/could be change/adjusted in order to maintain stability (it is after all a closed control loop): hopefully the circuit designer can tell/share with Don more about this behavior.

Will

 

[McGizmo]

Re: The Haiku: why the 3S McClickie lights have surpassed the PD lights!

Don, I was under the impression, from reading your post, that this correction was not happening as often as "once every few seconds." This is the relevant bit:



i.e. when it is correcting, it adjusts a few times, with a heart-beat type pulse, each "beat" being a 1/3 of a second, but then it goes back to a straight line for however long. It's not like it's beating/pulsing every few seconds indefinitely, right?

Please clarify, if you would.

js,
We can all speculate on what is going on and some of you guys can do a more informed job of speculation than I but I'm not sure what that accomplishes. I certainly don't want to make a big deal out of this recently identified "artifact" for likely obvious reasons; namely because I don't think the artifact is much of a deal but more a curiosity.

To answer your question though, I graphed the luminous output over time and from this we can guess what might be going on and see a representation of what light, in terms of magnitude, was coming out of the flashlight over the time period graphed. To make it easier, below is the graph itself:

The y axis is lumens measured and the x axis represents scans which were about 50 milliseconds apart from each other so I calculate the graph represents a little under a minute and a half in duration.

It is my understanding that the converter checks the current quite often and will not attempt a correction until the current is 10 mA above the set target level. On the very first fireup of the light engine, the converter goes through a calibration phase and sets its initial output voltage as based on Vf specific to the LED in the circuit. I believe that by the time the converter establishes this "starting point" the LED has had time to raise in temperature from room temperature and therefore a starting Vf is lower than what it really is at room temperature. I say this because I believe this explains why you see the initial lumens down below 150 in the graph. Now as the LED heats up, its Vf drops and we see the rise in lumens due to a constant current being provided to it coupled with a dropping Vf. At some point the Vf drops to a level that the Vin allows the current to rise in excess of the 10 mA ceiling allowed. We see three cycles starting at about 50 on the x axis where the converter makes adjustment and then the lumen curve straightens out but slowly rises due to temp increase and at about point 290 the converter again has to step in due to the 10 mA ceiling being reached again. It looks like it takes the converter fourteen steps in correction to stop the run away and then the lumen curve straightens out again and continues on until the end of the measuring period. You can see that the lumen curve is on the rise and no doubt due to the current slowly rising once again. At some point down the road it is likely that the current will again exceed the 10 mA ceiling and another correction will take place.

I imagine that once the operating temp of the light stabilizes that no additional corrections would be evident as they would not be necessary.

If I am reading this story correctly it seems to me that the best way to minimize the evidence of artifact associated with current correction would be to lower the ceiling allowed for current above target as this would presumably reduce the amplitude from low to high during the correction period. But I make assumptions here and it is likely that the converter would have to step in more often to make corrections. On the other hand, if the ceiling were increased the converter would probably have to step in fewer times for corrections but would these corrections be greater in amplitude and be that much more obvious? :shrug:

If anything, taking all this into consideration increases my appreciation for what these converters do as well as the designers who come up with them. :bow:

Now if this artifact resulting from the LED heating up can be reduced with a programing change that would be cool but if it is what it is, so be it.

If the discussion here becomes the primary focus of this thread I apologize to js. I personally think the relative significance of this discovery is minimal but that's just my opinion and I leave it to you guys. I feel I have met my responsibility in bringing it up.

 

[js]

Re: The Haiku: why the 3S McClickie lights have surpassed the PD lights!

js,
We can all speculate on what is going on and some of you guys can do a more informed job of speculation than I but I'm not sure what that accomplishes. I certainly don't want to make a big deal out of this recently identified "artifact" for likely obvious reasons; namely because I don't think the artifact is much of a deal but more a curiosity.

Absolutely. I asked the question because I didn't want others who were reading this to think that the light was pulsing constantly every few seconds (unless that really was the case). As your graph shows, this isn't the case. Again, just wanted put things in context for the benefit of those reading this thread.

If the discussion here becomes the primary focus of this thread I apologize to js. I personally think the relative significance of this discovery is minimal but that's just my opinion and I leave it to you guys. I feel I have met my responsibility in bringing it up.

No worries! I'm happy to have ANY discussion (excepting flame wars) going on in my threads. I think that given the topic of this thread (3S vs PD lights, compare/contrast, pros/cons) this particular subject fits right in and everyone should feel free to discuss this as they wish.

However, again, I completely agree with the sentiment that this artifact is what it is, a fairly minor quirk on the Haiku/SunDrop/3S lights, taken for all in all. So, if we return to our previous discussions, that would be great.

In particular . . . I'd love to hear you (Don) talk about Titanium some more! I was checking out that Haiku from the ashes threads, and it's just a real testament to Titanium as a material that the head and body weren't fused together, and that the o-ring channels were still in spec, and that the light still looked pretty cool with only some minor cleaning. And of course, you could have bead blasted the light or cleaned it until it looked like new, but it's better the way it is. But this is the point of Titanium, for me: durability and longevity! There's no reason that you couldn't keep your Haiku for a lifetime, maybe occasionally upgrading the LED or converter, but keeping the same host body for decades. That is truly a green material, a green light!

 

[scout24]

Re: The Haiku: why the 3S McClickie lights have surpassed the PD lights!

Keep the Haiku for a lifetime, indeed. I posted elsewhere that I thought one of it's greatest merits was being able to take down to component parts with a Leatherman tool. O-rings, switch, boot, window, LE can, reflector. With a modest amount of mechanical skill, it's the ultimate in user serviceable. I simplify, but it speaks volumes to me about the utility of the design and the thought that Don put into it that we're on the fourth generation of Haiku if you will (XRE, XPG, Nichia 119, XML) with no structural changes in the light. McClicky pack is timeless, and the flexibility of the head to produce the quality of beam offered with the reflectors that have to fit is fantastic. Being able to have one LE can and able to purchase different heads for different uses is a huge plus as well. Mule and Sundrop XRU head will host all current LE offerings for the Haiku, IIRC, allowing you to completely change the personality of your light in less time than it takes to type this. I can't think of anything else that offers this level of versatility, anywhere...


The PD's are certainly updateable by those with the requisite skill and patience, (I enjoy my neutral XML McLuxIII PD) but it's not like swapping LE cans.

 

[js]

Re: The Haiku: why the 3S McClickie lights have surpassed the PD lights!

Nice post, scout24! Quite a thought indeed--that we're on to the fourth generation of Haiku with no structural changes in the light. And the modularity of Don's current line up is also very, very cool.

 

[fyrstormer]

Re: The Haiku: why the 3S McClickie lights have surpassed the PD lights!

My apologies, I guess I assumed the adjustment was happening frequently or else the user who reported it wouldn't have cared.

 

[js]

Re: The Haiku: why the 3S McClickie lights have surpassed the PD lights!

I added Don's original Haiku thread to the top of the list of threads. Really good info there, especially if you want to know o-ring sizes, etc. Plus, hey, it's Don's own thread about the Haiku. What could be better?

 

[nbp]

I guess I missed that one in the list of links. Oops.

 

[arewethereyetdad]

Re: The Haiku: why the 3S McClickie lights have surpassed the PD lights!

Thanks for NOTHIN', JS. Now, because of you, I have to fork another $450 over to Don for another Haiku, a gem sorely missing from my collection. You have made me fall back in love with the Haiku, even though I swore from Day 1 that all I ever really needed were PD-based titanium McGizmos. Geez.......

(note to self: do not read js posts - they'll cost ya)

 

[wquiles]

Re: The Haiku: why the 3S McClickie lights have surpassed the PD lights!

(note to self: do not read js posts - they'll cost ya)

I hear you loud and clear. I have a McGizmo PD mule head with a clicky tailcap using a hi-CRI LED, a Saphire, two mechanical keyboards, etc.. - a lot of it due to "advice" and reviews from js

 

[js]

Re: The Haiku: why the 3S McClickie lights have surpassed the PD lights!

Oops! Ah . . . sorry guys!?!

But honestly, the Haiku is a freaking GEM of a light! I just LOVE LOVE LOVE this thing! I especially love the High CRI Nichia 119 flavor that I have. It's so nice to have such incredibly high quality light from my EDC again. So nice to come back to this kind of light. Of course, it was so nice to switch from the A2 to the LS20, i.e. to a Titanium PD light. The very cool light and lower CRI from the LS20 was a fairly small price to pay for all the benefits. But now? Now, well, I have a Titanium flashlight AND super high quality light all in one! Plus, the Haiku is just a work of perfection. It feels so great in the hand; it's so well made. I just can't say enough good things about this light.

Don't get me wrong, though, I still love my LunaSol 20. I was actually thinking of selling it for a while, thinking that if it wasn't my EDC that I should sell it and put the money towards more stuff for my house, etc. But then when it came to it, I was like, ah, no, sorry. I've carried this light almost every day for almost four years. There's no way I'm parting with it. I figure I'll carry it from time to time, just because I still love it so much.

I was reading in the Classic PD thread (link that Don posted, and which I added to list in OP) where someone was saying "Never sell a McGizmo." I think that's pretty damn good advice--I mean, obviously if you buy one and just decide it isn't for you, then that's one thing. But if you have it for a while and like it (or, more probably, if you love it) then DON'T sell it! You'll probably regret it. I sold my Ti-PD-S from the last wave of Ti-PD-S's. Yes, I had one of those for a couple months. When I bought it I planned from the beginning on selling it because I bought it for my LS20 review thread. I really liked it, though. What an awesome light! But I liked the LS20 more, and I felt I couldn't "justify" keeping both of them. So I sold it.

But now I wish I hadn't! It IS a classic!

However, all of that said, if I had to pick only one light to keep, it'd be my Haiku. Such a nice light!

So, for all of you people out there wishing for a Ti PD light of some sort or another, consider giving the Haiku a try. It's an amazing light, and just like one of those threads said in the title, I love it more and more. And so, probably, will you!

 

[pjandyho]

Re: The Haiku: why the 3S McClickie lights have surpassed the PD lights!

You are absolutely right js. No way would I ever sell my McGizmo. The Haiku and Mule are such nice lights I have been EDCing them everyday since I got them. Not a single day passed by without at least a Haiku in my pocket.

 

[nbp]

Right on Andy! I have been EDCing a Haiku and a Mule together since early last summer and I honestly think it's the finest combo out there. I do at times have to carry other smallish lights due to reduced pocket space, but for work, play, and general bombing around the Haiku and Mule pair always seem to end up clipped to my jeans together.

 

[pjandyho]

Re: The Haiku: why the 3S McClickie lights have surpassed the PD lights!

Right on Andy! I have been EDCing a Haiku and a Mule together since early last summer and I honestly think it's the finest combo out there. I do at times have to carry other smallish lights due to reduced pocket space, but for work, play, and general bombing around the Haiku and Mule pair always seem to end up clipped to my jeans together.

It's such a wonderful addiction eh? No regrets at all for making that move to splurge on McGizmo. My only regret is not making that plunge long ago when I should have.

 

[kaichu dento]

Re: The Haiku: why the 3S McClickie lights have surpassed the PD lights!

...the Haiku and Mule pair always seem to end up clipped to my jeans together.

Got a picture of that?

 

[Child of Rawls]

JS, I couldn't agree more. The 3S mclicky on the haiku is just amazing. IMO it makes all of these control ring set ups unnecessary. Three well spaced outputs with a great clicky, is pretty hard to beat. Thanks for putting into words what I have believed for so long.